How to tweak the code?

[thecoqa]

Hello, this might be a stupid question for all, but keep in mind that I'm a beginner in the software development scene, and any help would be amazing.
I wish to test the efficiency of the RECOGNITION feature and be able to tweak the code to manipulate the accuracy of said feature. Is there any way I can achieve that? (sorry if this question is too broad, just a little lost here). I don't want to improve it, but to test how little changes in the code affect a face being recognized, and if those changes have bigger impacts of some particular groups of people (age, gender, race)
I will be using Python SDK if that information helps.
Thanks in advance.